This is my Keras implementation of prostate MR images segmentation with Attention UNet(UNet with attention gates to make training foucus on Region of Interest)
Tested on Promise12 dataset, got DSC of 0.812
-
Download the promise12 dataset: download
know about the competition: Promise12 -
Put the training data in the folder data/train/ and unzip
-
Run the main code at codes/main.py
- Python3
- Required packages:
- Keras 2.2.4
- SKlearn
- scipy
- scikit image
- matplotlib
- Python-OpenCV
You can manage your python environment easily with Anaconda, and you can find the required packages with PYPI
Thanks to Nabsabraham and her focal-tversky-unet !